OBJECTIVES

We aim to estimate EQ-5D health state utilities for health states commonly used in oncology economic models informing HTA assessments based on 2 mature phase 3 studies involving nivolumab treatment in 2L non-small cell lung cancer (NSCLC).

METHODS

Utility scores were developed by applying Canadian weights to EQ-5D-3L data from pooled 5-year database locks of trials evaluating nivolumab versus docetaxel in patients with pretreated metastatic squamous (CheckMate-017) and non-squamous NSCLC (CheckMate-057). Repeated-measures mixed-effects models were used to estimate utilities. Health states were defined based on RECIST progression status categories, time to death (TTD) categories (≤4 weeks, 5-26 weeks, 27-52 weeks, or >52 weeks), and combinations thereof. Below we report on utilities for the overall sample. Additional models including randomized treatment and interactions with health state and TTD categories will be included in the presentation.

RESULTS

Overall across randomized treatments, 4 sets of health state utilities were estimated. The means with 95% CIs are:

I) Pre-progression: 0.751 (0.740, 0.762), post-progression: 0.697 (0.684, 0.711); no difference by treatment.

II) TTD (≤4 weeks: 0.560 (0.532, 0.589), 5-26 weeks: 0.701 (0.688, 0.713), 27-52 weeks: 0.758 (0.745, 0.771), >52 weeks: 0.795 (0.782, 0.808).

III) TTD (≤4 weeks: 0.656 (0.617, 0.694), 5-26 weeks: 0.722 (0.707, 0.736), 27-52 weeks: 0.756 (0.742, 0.770), >52 weeks: 0.785 (0.772, 0.798) and progressive disease decrement: -0.069 (-0.085, -0.053).

IV) Pre-progression: 0.769 (0.756, 0.782), post-progression: 0.760 (0.742, 0.778) and TTD ≤ 52 weeks decrement: -0.062 (-0.075, -0.048).

CONCLUSIONS

We used a systematic approach to estimate utilities for alternative health state definitions accepted by HTA bodies. The results document diminishing utility for NSCLC patients with progressed disease and approaching time to death. Progression seems to have significant impact on utility after taking into account time to death and vice versa.